Linux Kernel Community Drops Mandatory AI Tool Disclosure Requirements
Linus Torvalds rejects AI labeling, prioritizing code comprehension over the tools used for development.
1Policy Shift on AI
The Linux kernel community has officially decided to deprecate the mandatory Assisted-by tag for code contributions. This policy, which required developers to disclose the use of AI tools in their patches, was introduced in January 2026. Maintainers reached this consensus recently, signaling a significant shift in how the project manages automated coding assistance.
2Torvalds Rejects Mandatory Labels
Linus Torvalds has strongly opposed the disclosure requirement, labeling it as an idiotic concept. He compared the demand for AI transparency to absurd requirements like banning compilers or text editors. For Torvalds, the focus remains entirely on whether a contributor truly understands the code they submit, regardless of the tools used.
3Evolution of AI Quality
The decision follows a notable improvement in the quality of code generated by LLMs over the past year. Veteran maintainer Greg Kroah-Hartman noted at the Open Source Summit that AI models are now effective at identifying bugs. Because code relies on specific patterns, these models have become increasingly capable of producing acceptable kernel contributions.
4Arguments Against the Tag
Developer Jeff Layton highlighted several reasons for reversing the policy, including poor compliance rates among contributors. He argued that the tagging requirement functioned primarily as free advertising for proprietary LLM companies. Furthermore, he questioned the fundamental utility of collecting such data, suggesting it provided no tangible benefit to the kernel maintenance process.
5Internal Dissent Remains
Not all contributors support the removal of the disclosure requirement. Laurent Pinchart argued that the tag served as a mechanism for setting community expectations regarding code quality. He suggested that even if some developers ignore the rules, the policy provided a necessary framework for defining what the community accepts and rejects in submissions.
6Concerns Over Accountability
Despite the move toward acceptance, the community remains cautious about potential risks. Concerns persist regarding errors slipping through unnoticed, the possibility of AI agents reproducing copyrighted material, and the overall maintainability of software. There is also a lingering stigma in some circles against developers who rely heavily on AI tools for their work.
7Industry Attribution Discrepancies
The broader software industry lacks a unified approach to AI attribution. While companies like Anthropic include provenance messages by default, others like Codex do not. This inconsistency complicates efforts to track human creative input versus machine generation, leaving the Linux kernel community to prioritize functional code quality over standardized provenance tracking methods.
8Focus on Competence
The removal of the tag reflects a broader philosophy that intelligence is demonstrated by the ability to utilize new tools effectively. By rejecting restrictive rules, the Linux community aims to focus on the final output rather than the process. The core expectation remains that developers must fully understand and verify every contribution they submit.